BETA
BETA

More Numbers Behind Leo's Legendary Comebacks

By Jack "Jackie Peanuts" Moore | 01/04/22

Our first longform feature, The Legend of Game 4 MkLeo, premiered on our YouTube channel this weekend. It revolved around this massive chart, plotting MkLeo's
Byleth
Joker
set win rate at major tournaments after going down 0-2 in a Best-of-5 set:

Click to expand. It's almost literally in 4K, so you'll have to.
Today, I want to highlight one specific section of this graph that didn't get much coverage in the video itself. To do that, let's remove some of the clutter:
Click to expand.
These other lines, the red, green, and non-PGstats blue, represent the three players who have come closer than any other to replicating Leo's legendary ability to reverse 3-0. They are Shuton
Olimar
in red, Cosmos
Inkling
Corrin
Pyra/Mythra
in green, and Tweek
Diddy Kong
Wario
Pokémon Trainer
in blue.
Of players with at least eight reverse 3-0s in our database, Shuton and Tweek were the only other players besides Leo we could find to ever have a 50% or better reverse 3-0 rate at any time in their Ultimate careers -- yes, that means nobody else went even 1-1 or 2-2 in their first two or four chances at the reverse 3-0. A few points are marked on the chart; I'll refer to these later in the article.
Let's focus on the red and blue lines, first. Tweek and Shuton both lost their first sets on record going down 3-0. For Shuton, it was against Tweek himself at the first PGRU S-Tier, Genesis 6, where Tweek's Wolf sent Shuton home at 13th via a clean 3-0. Tweek would go on to lose his next set 3-0 to Dabuz
Olimar
Rosalina & Luma
, putting both Tweek and Shuton at 0-1 to begin Ultimate in these situations.
Here, we reach Point A. Frostbite 2019 remains one of the best events ever for both Tweek and Shuton, with each picking up a reverse 3-0 as part of their huge runs. Shuton dropped the first two games of his Top 8 Losers Round 1 set against Nairo
Palutena
before rallying to win the set; he would end up at 4th, still his best run at an American S-Tier.
A few sets later, Tweek found himself up against MkLeo for the first time Ultimate in Winners Finals, and after his Donkey Kong failed to do anything against MkLeo's Ike, the Wario lost game two and then proceeded to not just reverse 3-0 MkLeo but triple two-stock him.
Tweek wouldn't complete another recorded reverse 3-0 for nearly a full year. After falling nine straight times after going down 0-2, at Glitch 8's Winners Finals, Tweek was able to turn the tides against Dark Wizzy, leaving his pre-quarantine reverse 3-0 success rate at 14.2 (2/14).
Shuton, on the other hand, would also make the reverse 3-0 in his third opportunity at another American major, this time against his countryman Tea
Pac-Man
. This would be his peak, represented at Point B. Shuton would immediately fall to another compatriot, Zackray
Wolf
, 3-0, bumping his reverse 3-0 success rate down to 50%. At least in our database, Shuton has never done it at a major since then, failing in all six of his subsequent reverse 3-0 chances. He ended the pre-quarantine period with a reverse 3-0 rate of 20% -- still well above average; still well below half.
None of this is meant to downplay the accomplishments or abilities of Tweek and Shuton, but rather to show just how far MkLeo is from even his closest contenders. Cosmos, in green, is another great example. Cosmos is one of just three active players to reach three recorded reverse 3-0s prior to quarantine, joining Leo and Tweek. He has been one of the clutchest players in the game. His reverse 3-0s are represented by Points C.
And yet he is nowhere near Leo's level by this metric. His peak reverse 3-0 rate came after he achieved his third recorded major reverse 3-0, in Losers Quarterfinals against Dabuz at Glitch 8 in January 2020, where he switched from Inkling to Corrin to pull the reverse sweep. That put Cosmos's recorded reverse 3-0 rate at 23%, the highest it would ever get.
With Leo's stats, even though we don't have game-by-game data for every single Best-of-5 he's played, even at PGRU events, we were able to cobble together what little Smash.gg was missing via VoDs. Leo really doesn't go down 0-2 that often, and when he does, it's streamed. That's not the case with the rest of these players. At most offline tournaments, Best-of-5 sets begin well before Top 8, but streamed matches are often the only ones where data is recorded, which typically misses a huge portion of losers bracket. So I accounted for that with the huge peaks at the end of the graph, represented at Points D.
We do have the game counts for all of their best-of-5 sets in the database, so we can at least look at all 3-2 sets, which are the only ones that could possibly have been reverse 3-0s. Tweek has won eight other sets by a 3-2 count; Shuton has won seven; Cosmos has won nine. Even if we were to assume each and every single one of these 24 sets were reverse 3-0s, these three players would all still have reverse 3-0 success rates worse than Leo's current 57.8% mark:
Cosmos: 3/19 recorded -> 12/28 = 42.9%
Tweek: 2/14 recorded -> 10/22 = 45.4%
Shuton: 2/10 recorded -> 9/17 = 52.9%
Even with the most generous possible assumptions, Shuton remains about five percent shy of Leo's legendary level; the other two don't even break the 50 percent mark.
Another way to put this into perspective: to reach the measured average rate of reverse 3-0s prior to the pandemic, 11.7 percent, MkLeo would need to lose his next 75 sets in a row when going down 0-2.
No matter how the rest of Ultimate's competitive history writes itself, MkLeo's reverse 3-0s will always stand out as the most eye-popping phenomenon in the game, and possibly in esports at large. Whatever the game count, whoever you are: If you're playing against MkLeo, it's your back that's against the wall.
Did you miss the premiere of PGstats's first longform video, The Legend of Game 4 MkLeo, this weekend? Check it out now, and subscribe to the PGstats YouTube today:
Jackie Peanuts is the Director of PGstats. A sportswriter and statistician, Jackie's work has appeared on ESPN, CBS, Vice, The Guardian, The Atlantic, and others.